融合定位算法的YOLO-V4模型实现电力杆塔状态评估  

State Evaluation of Power Pole Tower Based on YOLO-V4 Model with Fusion Positioning Algorithm

作　　者：阮远峰 郭建武 蔡金涛 RUAN Yuanfeng;GUO Jianwu;CAI Jintao(State Grid Zhejiang Electric Power Co.,Ltd.,Deqing County Power Supply Co.,Deqing 313200,China)

机构地区：[1]国网浙江省电力有限公司德清县供电公司,浙江德清313200

出　　处：《电工技术》2023年第13期32-34,40,共4页Electric Engineering

摘　　要：针对电网线路在使用中经常出现断杆、倒杆现象,而现有的技术性检测方式已不能满足要求的问题,给出了一种新的电力杆塔状态评估方法,并建立了改善的YOLO-V4模型,以实现电力杆塔的状态精准定位。CSPDarknet-53模型试验说明,该方法能精准定位电力杆塔故障点,偏差小。The phenomenon of broken rod and inverted rod often occurs in the process of using power grid lines,and the existing technical detection methods can no longer meet the requirements.To solve this problem,a novel power tower condition evaluation method was given,and an improved YOLO-V4 model was established to achieve accurate state positioning of power towers.The CSPDarknet-53 model test showed that the method could accurately locate the fault point of the power pole tower with small deviation.

关 键 词：电网线路 状态评估 YOLO-V4模型 CSPDarknet-53特征 目标检测 

分 类 号：TM73[电气工程—电力系统及自动化]